How to go from an SDTM Finding Domain to an ADaM-Compliant Basic Data Structure Analysis Dataset: An Example

Qian Wang, MSD, Brussels, BelgiumCarl Herremans, MSD, Brussels,

BelgiumPhUSE 2010, Berlin

PhUSE 2010 - Berlin 2

ADaM versus SDTM ADaM datasets do use redundancy for easy

analysis – common variables may be found across all analysis datasets (e.g., population flags, subject identifiers, etc.);

ADaM datasets have a greater number of numeric variables (e.g., SAS formatted dates, numeric representation of a character grouping variable from SDTM);

ADaM datasets may combine variables from multiple SDTM domains;

ADaM datasets are named AD<xxxxxx>.

PhUSE 2010 - Berlin 3

ADaM-compliantBasic Data Structure (BDS)

A BDS dataset contains one or more records per subject, per analysis parameter and per analysis time point.

It describes the data being analyzed and also includes variables to support the analysis (e.g. covariates) as well as information to facilitate traceability.

PhUSE 2010 - Berlin 4

6 Basic Steps

Step 1Prepare ADaM

Precursor

Step 2Convert ISO8601

Dates

Step 3Create

Analysis Relative

Day

Step 4Create Key

Analysis Variables

Step 5Add

Time Windows

Step 6Derive Change

from Baseline

PhUSE 2010 - Berlin 5

STEP 1: Preparing for ADAM Precursor by Adding SUPPQUAL to SDTM

LB

SUPPQUAL

LB + SUPPQUAL (where RDOMAIN=‘LB’)

PhUSE 2010 - Berlin 6

STEP 1: Preparing for ADAM Precursor by adding SUPPQUAL to SDTM

USUBJID SEX LBSEQ LBTESTCD LBORRES LBFAST

0002_0001 M 1 NA 0.1 Y

STUDYID

RDOMAIN

USUBJID IDVAR IDVARVARL

QNAM QVAL

0002 LB 0002_0001 LBSEQ 1 LBFAST Y

USUBJID SEX LBSEQ LBTESTCD LBORRES

0002_0001 M 1 NA 0.1

+

PhUSE 2010 - Berlin 7

STEP 1: Preparing for ADAM Precursor by Adding SUPPQUAL to SDTM

proc transpose data=supplb out=supplb2

(rename=(idvarval=lbseq)); by usubjid idvarval; var qval; id qnam; idlabel qlabel;run; %addsuppqual(input_libary =,

input_dataset_suppqual =,input_dataset_define =,

output_library = );

PhUSE 2010 - Berlin 8

6 Basic Steps

Step 1Prepare ADaM

Precursor

Step 2Convert ISO8601

Dates

Step 3Create

Analysis Relative

Day

Step 4Create Key

Analysis Variables

Step 5Add

Time Windows

Step 6Derive Change

from Baseline

PhUSE 2010 - Berlin 9

STEP 2: Converting ISO8601 Date/Time

--DTC

2003-12-15T13:14:17

PhUSE 2010 - Berlin 10

STEP 2: Converting ISO8601 Date/Time

If not missing ADT and ATM populate ADTM

PhUSE 2010 - Berlin 11

STEP 2: Converting ISO8601 Date/Time

--DTC ADT ADTF ATM ATMF ADTM

2003-12-15T13:14:17 15Dec2003 13:14:17 15Dec2003:13:14:17

2003-12-15T13:14 15Dec2003 13:14:00 S 15Dec2003:13:14:00

2003-12-15T13 15Dec2003 13:00:00 M 15Dec2003:13:00:00

2003-12-15 15Dec2003

2003-12 01Dec2003 D

2003 01Jul2003 M

%convert2date(input_dataset=, output_dataset=

);

PhUSE 2010 - Berlin 12

6 Basic Steps

Step 1Prepare ADaM

Precursor

Step 2Convert ISO8601

Dates

Step 3Create

Analysis Relative

Day

Step 4Create Key

Analysis Variables

Step 5Add

Time Windows

Step 6Derive Change

from Baseline

PhUSE 2010 - Berlin 13

STEP 3: Creating Analysis Relative Day

ADY = LBDT - ADSL. TRTSDT + (LBDT > ADSL.TRTSDT)

PhUSE 2010 - Berlin 14

6 Basic Steps

Step 1Prepare ADaM

Precursor

Step 2Convert ISO8601

Dates

Step 3Create

Analysis Relative

Day

Step 4Create Key

Analysis Variables

Step 5Add

Time Windows

Step 6Derive Change

from Baseline

PhUSE 2010 - Berlin 15

SDTM LBLBTESTCD LBTEST LBSTRESN LBSTRESU

NA Sodium 139 mmol/L

K Potassium 3.5 mmol/L

ADaM ADLB

PARAMCD PARAM AVAL

NA Sodium (mmol/L) 139

K Potassium (mmol/L) 3.5

STEP 4: Creating Key Analysis Parameter Variables PARAM(CD), AVAL

PhUSE 2010 - Berlin 16

STEP 4: Creating Key Analysis Parameter Variables PARAM(CD), AVAL

SDTM LB

LBDY LBTESTCD LBTEST LBSTRESN LBSTRESU

1 NA Sodium 139 mmol/L

1 NA Sodium 145 mmol/L

ADaM ADLB

ADY PARAMCD PARAM AVAL DTYPE ANL01FL

1 NA Sodium (mmol/L) 139

1 NA Sodium (mmol/L) 145

1 NA Sodium (mmol/L) 142 AVERAGE Y

PhUSE 2010 - Berlin 17

STEP 4: Creating Key Analysis Parameter Variables PARAM(CD), AVAL

%finding(input_dataset=, domain=, summaryfunction=AVERAGE, output_dataset= );

PhUSE 2010 - Berlin 18

6 Basic Steps

Step 1Prepare ADaM

Precursor

Step 2Convert ISO8601

Dates

Step 3Create

Analysis Relative

Day

Step 4Create Key

Analysis Variables

Step 5Add

Time Windows

Step 6Derive Change

from Baseline

PhUSE 2010 - Berlin 19

Time Point (Target Day)Time window

in Days

Baseline (1) -70 to 7

Treatment 1 (45) 8 to 90

Treatment 2 (135) 91 to 180

1

Time

-70 7 90 180

Baseline Treatment 1 Treatment 2

STEP 5: ADDING TIME WINDOWS

PhUSE 2010 - Berlin 20

STEP 5: ADDING TIME WINDOWS

Start day Stop day Target AVISIT AVISITN Baseline

-70 7 1 Baseline 0 Y

8 90 45 Treatment 1 1

91 180 135 Treatment 2 2

ADY PARAMCD AVAL AVISIT AVISITN AWTARGET AWRANGE ABLFL ANL02FL

-18 NA 141 Baseline 0 1 -70 to 7

1 NA 140 Baseline 0 1 -70 to 7 Y Y

14 NA 145 Treatment 1 1 45 8 to 90

46 NA 149 Treatment 1 1 45 8 to 90 Y

PhUSE 2010 - Berlin 21

6 Basic Steps

Step 1Prepare ADaM

Precursor

Step 2Convert ISO8601

Dates

Step 3Create

Analysis Relative

Day

Step 4Create Key

Analysis Variables

Step 5Add

Time Windows

Step 6Derive Change

from Baseline

PhUSE 2010 - Berlin 22

Step 6: Deriving Change From Baseline

ADY PARAMCD AVAL AVISIT AVISITN ABLFL BASE CHG

-18 NA 141 Baseline 0 140 1

1 NA 140 Baseline 0 Y 140 0

14 NA 145 Treatment 1 1 140 5

46 NA 149 Treatment 1 1 140 9

%change(input_dataset=, input_timewindow_dataset=, output_dataset= );

PhUSE 2010 - Berlin 23

Conclusion

Step 1Prepare ADaM

Precursor

Step 2Convert ISO8601

Dates

Step 3Create

Analysis Relative

Day

Step 4Create Key

Analysis Variables

Step 5Add

Time Windows

Step 6Derive Change

from Baseline

6 Steps have been described to transform a SDTM finding dataset into an ADaM-compliant BDS analysis dataset.